Optimal Questionnaire Design for Student Assessment: A New Technique based on Item Response Theory

Along with the spread of web-based learning systems, there is the growing need for fast and accurate techniques for student assessment. In this field, the use of on-line questionnaires - based on multiple-choice items - is one of the most widespread approaches. This paper presents a new technique for adaptive questionnaire design by automatic selection of multiple-choice items. The approach uses a genetic algorithm to define the optimal set of items for questionnaire design for a given set of students, according to the Item Response Theory. The experimental results, carried out on both simulated and genuine data, confirm the effectiveness of the new approach for adaptive questionnaire design.